Coding expertise is going to collapse from AI reliance • Hacker News Board

Re: Coding expertise is going to collapse from AI reliance

#551

Post by monkaiju » Tue, Aug 25, 2026, 2:50 PM UTC

mekael wrote: post #483

monkaiju wrote: post #471

Earlier quoted context omitted.

Where I work we actively ban LLM use in development. We're hiring too so if you're willing to live in Salt Lake City...

May i ask what industry youre in? Also, how’s the winter in SLC?

We are a franchisor the owns 2 concepts, both of which are thrift stores aimed at different demographics. We write all of our software in-house and its actually a surprisingly "fun" context.

The business had pretty questionable (read: bad) software for a long time, then came under new management a few years ago. I joined just under two years ago and since then we've been modernizing the code with an eye to move it off of the in-store servers its running on currently and into the cloud. That is complicated by the fact that the franchisees are not used to certain things, like really needing to have reliable internet with a failover, so we've had to (or had the opportunity) do some serious hardening on the in-store servers and how we deploy software to them. We've switched them all to immutable OSes, added a bunch of telemetry, and added some new services that are slowly eating the old ones' responsibilities and will be much easier to move to the cloud.

Also the winters are ok but not as nice as they used to be. I grew up here so I remember when they were extremely snowy and I miss that, now they're fairly grey and mild.

Re: Coding expertise is going to collapse from AI reliance

#552

Post by GPerson » Tue, Aug 25, 2026, 3:05 PM UTC

zqna wrote: post #435

GPerson wrote: post #328

Earlier quoted context omitted.

I’m assuming you’re a software engineer. The AI programming meta that keeps evolving (last month it was loops) seems like a relatively trivial skill to adopt compared to learning what it’s like to do high quality work and to participate in a team, and these kinds of things. If the AI stops improving then everyone will master the AI skill set that everyone else is using, and people will acquire maturity and experience…

There 4 categories of people: the bandits, the smarts, the naives and the stupids. Bandits reign over the world, and in the past they needed the naives and the stupids to make them rich. In the age when enrichment can be automated they can see that naives and stupides are only contributing in shrinking their cake. They only need few smarts to keep their infrastructure alive so they could enjoy the planet for themselv…

The world consists of independent actors, at different scales, and lots of entropy. Any model like yours giving all the agency to 50 protagonists is a silly oversimplification.

Re: Coding expertise is going to collapse from AI reliance

#553

Post by peab » Tue, Aug 25, 2026, 3:16 PM UTC

ryandvm wrote: post #48

100% We're already seeing this at the enterprise level. Companies have dictates from leadership that "if you're writing code manually, you're doing it wrong." Okay, that kind of works for a while. We are indeed producing a shit-ton of code, but the reality is that engineers are pumping out code faster than the humans can understand and (honestly) review it. That sounds great until you realize that "hey Claude, read t…

Would you say the same thing about people not knowing assembly after compilers came out?

Re: Coding expertise is going to collapse from AI reliance

#554

Post by docmars » Tue, Aug 25, 2026, 4:49 PM UTC

apatheticonion wrote: post #480

larsfaye wrote: post #445

Earlier quoted context omitted.

This is EXACTLY how I am working with these tools, you summarized everything about why I use this approach perfectly. Zed + OpenCode + Mimo/GLM/Kimi models are incredible. I haven't touched a "frontier model" for months. I really treat them as "delegation layer". They're systems that sit right alongside your stack, and at this point, at every level in it, like a connective tissue that intersects at all points. I can…

I had a job interview at Canva where they had an "AI coding round". They wanted to see how I used AI in my workflow. Because I was using DeepSeek flash, the interviewer was confused at how fast I was getting edits - he'd only used Claude and was used to it thinking for 40 minutes to update a function. We were building a frontend application and I was able to do that whole guided prompt/review loop. I finished the tas…

So frustrating. Too many hiring teams seeking the "perfect" candidate instead of hiring for completely valid and amazing potential.

I'd wager you're far better than whatever candidate they landed on, because the person who apparently qualified for that role probably isn't thinking while they prompt away.

Why are engineering teams looking for brainless monkeys? It's to their detriment all around if they aren't hiring capable people.

Re: Coding expertise is going to collapse from AI reliance

#555

Post by wry_discontent » Tue, Aug 25, 2026, 4:55 PM UTC

hintymad wrote: post #367

wry_discontent wrote: post #278

Earlier quoted context omitted.

They work on research problems you can define a clear solution criteria for. The kinds of business software I work on don't have those characteristics. If I needed something like a utils library, I think I could easily have Claude write the whole thing and not read the code.

What puzzles me is this: research means that we are exploring something that has not done before, yet using Claude to generate code means slicing and dicing what has been done many times before. So, I'm not sure how to make sense of both at the same time: Anthropic is pushing the boundary of AI, yet all the knowledge and engineering in form of code can be generated from the previous work?

How I interpreted their thoughts on "loops" and "graphs" etc, is that they 1. have unlimited token allowances, and 2. are working on problems where the solution itself can be described by code.

It reminds me of property based testing. I really liked the idea when I heard about it, but most of what I work on does not tend to behave in a way that can be easily described mathematically like that.

Maybe I'm just not clever enough to do it.

Re: Coding expertise is going to collapse from AI reliance

#556

Post by chermi » Tue, Aug 25, 2026, 5:19 PM UTC

latentsea wrote: post #515

chermi wrote: post #233

Earlier quoted context omitted.

This lands. Llms are bad precisely at following what not to do. They work best off of positive constraints. I have a design principles + tech preferences doc I force llms "lint" their approach against. It's not perfect but it helps. I call it a bias field, pushes them toward hopefully the happy and harmonious (with the rest of the system) paths. Obviously this is only partial and imperfect enforcement, but if it's ap…

>This lands. Where does it land?

Did I use the saying wrong? My bad I confuse them sometimes

Re: Coding expertise is going to collapse from AI reliance

#557

Post by matsemann » Tue, Aug 25, 2026, 6:27 PM UTC

bogdanoff_2 wrote: post #550

matsemann wrote: post #532

Earlier quoted context omitted.

Sure, but I'm only able to do that because I'm intimately aware of everything in the code base. The more I do this where the LLM ends up writing the code (even based on my decisions), the less good critique I'll be able to give in the future.

Is that the experience you had? What if you ask it questions about the codebase? I find AI agents really great for codebase exploration and understanding how it works. In some ways even moreso than from manual implementation since it's easier to get a global picture. You can also ask it questions like "are there recurring patterns of how X is done in the codebase?" to which it might answer sometime like "there are ac…

Not everything is encoded in code, though.

Re: Coding expertise is going to collapse from AI reliance

#558

Post by ksd482 » Tue, Aug 25, 2026, 6:33 PM UTC

ksd482 wrote: post #406

khriss wrote: post #362

Earlier quoted context omitted.

> I use AI to generate a textbook style reading material for me along with "verify yourself" types of exercises along the way This is fascinating . Any chance you have your workflow or prompts documented and would be willing to share them?

Great idea! Let me pull some stuff from my past chats, and put it in a github gist or something.

So I went back to my chats and there wasn't much there to put in a workflow or something. But here are the things that stood out:

> Introduce me to these concepts layer by layer with a textbook style introductions.

> Use color coded boxes etc. just like in textbook to highlight important details.

> Along the way, give me exercises to do myself that would test my understanding so far.

> Put everything in a local HTML file that I can open in my browser.

> The entire textbook should be printable so format accordingly.

That's about it.

The HTML part is important. Earlier I used to document in markdown format, but I found claude/chatgpt does a far better job with formatting and colors in HTML instead. Plus, you can open the HTML file in a browser and refresh when it makes changes.

Another thing: the HTML is built gradually. I keep asking AI some follow up questions and I instruct it to keep the HTML up todate.

Re: Coding expertise is going to collapse from AI reliance

#559

Post by ex1fm3ta » Tue, Aug 25, 2026, 6:49 PM UTC

lrvick wrote: post #226

ex1fm3ta wrote: post #189

Keep in mind also that LLMs are currenctly heavily subsidized. Once VCs money are gonna run out, you will see the real price, and your 200K per year dev is probably nothing.

A $1200 GPU is all a reasonably capable engineer needs for each work stream. The cloud prices are a scam.

hmm that's a bit low right now, since you need at least 48Gb of VRAM to be confortable. 32Gb might do it (on quantitized models + optimized) but it would be very slow.

Re: Coding expertise is going to collapse from AI reliance

#560

Post by tired_and_awake » Tue, Aug 25, 2026, 6:53 PM UTC

larsfaye wrote: post #36

tired_and_awake wrote: post #19

Earlier quoted context omitted.

This is exactly the right analogy. Granted LLMs are a more sophisticated calculator and engineering is still adapting to this wonky/evolving technology. But to pretend it's a coming catastrophe is to ignore every past parallel development - where this has not happened.

Math skills are at their lowest point ever in modern society, and many are still advocating for the reduction in reliance on calculators for this reason. https://ivyleaguecenter.org/2024/03/12/over-reliance-on-calc...

Would you advocate for the removal of calculators to fix this problem?